Improvement in NeuropathyOutcomes With NormalizingHbA1c in Patients With Type 2DiabetesDiabetes Care 2019;42:110–118 | https://doi.org/10.2337/dc18-1560

OBJECTIVE

To investigate the impact of normalizing HbA1c by extensive HbA1c control (EHC) onneuropathy outcome measures (NOMs), nephropathy, and retinopathy in type 2diabetes.

RESEARCH DESIGN AND METHODS

Detailed clinical and neurological examinations were performed in two cohortsof 38 patients with uncontrolled type 2 diabetes (HbA1c 9.6% [81.4 mmol/mol]) atbaseline and after glycemic control (GC) with or without EHC by diet restriction andhypoglycemicagentsover4yearsalongwith48control subjectswithnormalglucosetolerance (NGT) and 34 subjects with impaired glucose tolerance (IGT) only atbaseline.EHCpatients, controlsubjects,andsubjectswith IGTunderwentoral glucosetolerance tests. Glycemic variability (GV) was evaluated by SD and coefficient ofvariation of monthly measured HbA1c levels and casual plasma glucose.

RESULTS

In the EHC cohort, HbA1c levels over 4.3 years and the last 2 years improved to 6.1%(43.2 mmol/mol) and 5.8% (39.9 mmol/mol) with 7.3 kg body wt reduction, and50% and 28.9% of patients returned to IGT and NGT, respectively, at end point.Baseline neurophysiological and corneal nerve fiber (CNF) measures were im-paired in patients. Normalized HbA1c with EHC improved neurophysiological andCNF measures to be similar for those for IGT, while GC without EHC (mean HbA1c

level 7.0% [53.5 mmol/mol]) improved only vibration perception. The meannormalized HbA1c levels by EHC determined NOM improvements. The high GVand baseline HbA1c levels compromisedNOMs. Albumin excretion rate significantlydecreased, while retinopathy severity and frequency insignificantly worsened onEHC.

CONCLUSIONS

Normalizing HbA1c in type 2 diabetes of short duration improves microvascularcomplications including neuropathy and nephropathy more effectively than stan-dard GC but not retinopathy.

Intensive glycemic control (GC) has shown an equivocal efficacy regarding diabeticperipheral neuropathy (DPN) in type 2 diabetes (1), mainly due to nonoptimizedHbA1c levels. Randomized trials (2,3) have not been able to establish the opti-mum GC level for improving neuropathy outcomes in type 2 diabetes. In type 1

1Ishibashi Clinic, Hiroshima, Japan2University of ExeterMedical School, Exeter, U.K.

Corresponding author: Mitra Tavakoli,m.tavakoli@exeter.ac.uk

Received 20 July 2018 and accepted 12 October2018

This article contains Supplementary Data onlineat http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc18-1560/-/DC1.

© 2018 by the American Diabetes Association.Readers may use this article as long as the workis properly cited, the use is educational and notfor profit, and the work is not altered. More infor-mation is available at http://www.diabetesjournals.org/content/license.

Fukashi Ishibashi,1 Miki Taniguchi,1

Aiko Kosaka,1 Harumi Uetake,1 and

Mitra Tavakoli 2

110 Diabetes Care Volume 42, January 2019

PATH

OPHYSIOLO

GY/COMPLICATIONS

diabetes, near-normoglycemia for 24years prevented nerve function declines(4). The simultaneous pancreas-kidneytransplantation (SPK) can normalize HbA1clevels without severe hypoglycemia andwas shown to improve corneal nervefiber (CNF) measures (5). However, thelong-standing normalization of HbA1c lev-els by insulin-providing agents (IPAs) inpoorly controlled type 2 diabetes seemsinfeasible owing to potential severe hy-poglycemia (3). Therefore, the benefit ofnormalizing HbA1c levels without hypo-glycemia to neuropathy outcome mea-sures (NOMs) in type 2 diabetes has neverbeen investigated.Besides mean GC levels, glycemic var-

iability (GV) (6) and metabolic memoryinfluence DPN and other diabetic micro-angiopathies (7,8). Impaired glucose tol-erance (IGT) may cause neuropathy andalter CNF morphology as an indicator ofsmall-fiber neuropathy (9).Here, we aim to investigate the impact

of normalized HbA1c levels without hy-poglycemia on neurophysiological func-tions and CNF measures along withretinopathy and nephropathy in pa-tients with poorly controlled type 2diabetes.

RESEARCH DESIGN AND METHODS

SubjectsAmong 671 patients who newly vis-ited the Ishibashi Clinic between May2011 and July 2013 after the diagnosisof hyperglycemia, we extracted 38 pa-tients with type 2 diabetes fulfilling thefollowing criteria (extensive HbA1c con-trol [EHC] cohort): 1) baseline HbA1clevel .7.5% (58.5 mmol/mol) (meanlevel 9.6% [81.4 mmol/mol]), 2) fol-low-up period .3 years, and 3) annualmean HbA1c levels ,6.0% (42.1 mmol/mol) for the last 2 years before the finalvisit. Patients visited Ishibashi Clinic af-ter repeated diagnosis of hyperglycemia.Thirty-eight patients with type 2 diabetesmatched for age, sex, baseline HbA1clevels, and height to patients with EHCwere followed without EHC and servedas a control for EHC patients. Forty-eighthealthy control subjects with normalglucose tolerance (NGT) and 34 subjectswith IGT were enrolled and studied onlyat baseline.The patients with diabetes with or

without EHCwere followed for 3–5 years(4.3 and 4.1 years on average, respec-tively) until September 2017. Patients

who were treated by insulin-sensitizingagent (ISA) or IPA were given biguanidesor pioglitazone or given sulfonylureas orinsulin, respectively (10). The exclusioncriteria were as follows: any other clin-ically evident causes of neuropathyapart from diabetes, vitamin B defi-ciency, nonproliferative severe or pro-liferative diabetic retinopathy as definedby the Early Treatment Diabetic Reti-nopathy Study (ETDRS) (11), cornealdiseases, history of refractive surgery,and use of contact lenses. All subjectsprovided written informed consent asper the Declaration of Helsinki, and thestudy protocol was approved by theethics committee of Ishibashi Clinic.

Clinical and Laboratory DataPatients with diabetes with EHC at endpoint, control subjects, and subjects withIGT at baseline underwent a 75-g oralglucose tolerance test (OGTT). The BMI,blood pressure, casual postprandialplasma glucose (CPPG), and HbA1c levelswere measured monthly during theterms of study. In patients with type 2diabetes, SD and coefficient of variation(CV) of CPPG and HbA1c levels over thefollow-up period were calculated for esti-mating GV. The serum lipid levels andurinary creatinine and albumin levels wereassessed every 3 months. An albumin-to-creatinine ratio (ACR) .30 mg/g cre-atinine twice a year was labeled as ne-phropathy (12).

At baseline and end point, bilateralretinal fundus images (45°) were cap-tured and graded according to the ETDRSscale as no apparent retinopathy, 0; mildnonproliferative diabetic retinopathy, 1;and moderate nonproliferative diabeticretinopathy, 2 (11).

Assessment of NeuropathyAll patients with diabetes at baseline andend point, control subjects, and subjectswith IGT at baseline underwent detailedneurophysiological examinations. Neu-rological deficits were assessed usingthe modified neuropathy disability score(NDS) (13), which evaluates vibrationperception, pinprick, temperature per-ception, and ankle reflexes. The classifi-cation for evaluation of neuropathywas based on the recommendationsof the Toronto Consensus on Dia-betic Neuropathies (14), which considersa confirmed DPN as a combinationof the presence of abnormal nerve

conduction velocity (NCV) and a symp-tom, or a sign. As a result, patients withNDS .2 and sensory NCV (SCV) of suralnerve ,42 m/s were labeled with neu-ropathy based on the Toronto consen-sus. The cutoff level of sural nerve SCVwas determined according to themean 6 2 SD (49.99 6 8.02 m/s) ofsural nerve SCV in healthy control sub-jects (n = 38), matched for age, sex, andheight to patients with diabetes withEHC, recruited from 98 healthy controlsubjects at Ishibashi Clinic.

Electrophysiology and NCV studieshave been measured with an electromy-ography instrument (Neuropak S1;NihonKohden, Tokyo, Japan). The motor NCV(MCV) (mediannerve) andSCV (ulnar andsural nerves) and their action potentialamplitudes were determined. Skin tem-perature was maintained above 32°C.

The vibration perception threshold(VPT) was measured at the left medialmalleolus using a biothesiometer (Bio-Medical Instrument Company, Newbury,OH). Warm (WPT) and cold (CPT) per-ception thresholds at the dorsum of thefoot were determined using a thermalstimulator (Intercross-200; Intercross Co.,Tokyo, Japan). CAN was assessed by theCVof R-R intervals (CVR-R) calculated fromtheR-R intervalsof200electrocardiogramsamples.

Corneal Confocal MicroscopyAll patients with diabetes at baseline andend point, control subjects, and subjectswith IGT once at baseline were examinedusing a Heidelberg Retina Tomograph(HRT III) in vivo corneal confocal micro-scope with the Rostock Corneal Module(Heidelberg Engineering, Heidelberg,Germany) (15). Based on an establishedprotocol (16), six high-quality images persubject from Bowman’s layer were cap-tured and analyzed to quantify the fol-lowing CNF morphological parameters:1) CNF density (CNFD): total number ofmajor nerve fibers/mm2 corneal tissue,2) CNF length (CNFL): total length ofall nerve fibers (mm/mm2), 3) cornealnerve branch density (CNBD): numberof branches emanating from all majornerve trunks/mm2, 4) beading frequency(per 0.1mm), and5) bead size (mm2) (17).Except for bead size, all measure-ments were performed using ImageJ(Texelcraft, Tokyo, Japan). The exam-iners and the team members whoanalyzed the images and study results

care.diabetesjournals.org Ishibashi and Associates 111

were all blinded and masked to thestudy groups.

Statistical AnalysisAll statistical analyzes were performedusing SPSS, version 19 (SPSS, Chicago, IL),and P value ,0.05 was considered sta-tistically significant.A priori analysis of sample power (a =

0.05;b=0.80)usingG*Power3.1 (http://gpower.software.informer.com/3.1/) re-vealed that 53 subjects per each groupwere required. We could not recruitenough subjects. Thenapost hocanalysisof sample powerwas conductedwith useof a one-sided ANOVA (significance of0.05) and the Kruskal-Wallis test for CNFand neurophysiological measures. Thesubject’s statistical power ranged from0.89 to 0.99. Therefore, 158 subjects intotal sample size gave us adequate sta-tistical power.All values are presented as mean 6

SEM. All data sets were tested for nor-mality using the Shapiro-Wilk test. Thedifferences between baseline and endpoint in two patient cohorts were as-sessed using the paired t test and Wil-coxon signed rank test for normally andnonnormally distributed continuousvariables, respectively, and x2 test andMcNemar test for normally and nonnor-mally distributed categorical variables,respectively. Values at baseline in pa-tients with or without EHC and controlsubjects or between NGT and IGT sub-groups categorized by end point 75-gOGTT in patients with EHC were com-pared by t test and Mann-Whitney testfor normally and nonnormally distrib-uted continuous variables and x2 testfor categorical variables. The comparisonof normally distributed variables be-tween end point in two patient cohorts,control subjects and subjects with IGT,was performed using one-way ANOVAfor continuous variables and the x2 testfor categorical variables followed byBonferroni correction. For nonnormallydistributed variables, the Kruskal-Wallistest was applied followed by the Mann-Whitney U test and Bonferroni correctionfor continuous variables and the x2 testfor categorical variables. Correlationsbetween the changes in NOMs by GCand mean clinical factors over follow-upperiod, GV parameters, or baseline HbA1clevels in patients with EHC were as-sessed using Spearman rank correlationcoefficient or multiple regression analysis.

The sensitivity and specificity of CNFand neurophysiological measures in dif-ferentiating between control subjects,subjects with IGT, and patients withEHC were assessed at baseline and endpoint using receiver operating character-istic analysis.

RESULTS

Demographic DataThe sex, age, height, and baseline HbA1clevels were matched between two co-horts with diabetes with or without EHC.During follow-up period, BMI in patientswith EHC decreased (mean6 SEM bodyweight reduction 7.36 1.2 kg) but not inpatients without EHC. Systolic (SBP) anddiastolic (DBP) blood pressure at endpoint in two patient cohorts significantlydecreased after GC. In patients withEHC, HbA1c levels for the last 2 yearsof the follow-up period and at end pointwere ,6.0% (42.1 mmol/mol). Themean HbA1c levels during the wholefollow-up period in patients withoutEHC were higher than those in patientswith EHC. HbA1c levels in control subjectswere lower than those at end point intwo patient cohorts and in subjects withIGT. In patients with EHC, SD and CV ofCPPG and HbA1c levels over the last2 years from end point were signifi-cantly lower than those over the wholefollow-up period. Plasma glucose levelsbefore and after 75-g OGTT (30, 60,and 120 min) at end point in patientswith EHC were 102 6 2.2, 203 6 7.2,183 6 5.0, and 152 6 7.0 mg/dL, re-spectively. At end point, 120-min plasmaglucose in EHC and IGT subjects wassignificantly higher than that in controlsubjects (Table 1). Of patients on EHC,50.0% and 28.9% returned to IGT andNGT, respectively. Serum triglyceridelevels in patients with EHC decreasedafter GC. The end point estimated glo-merular filtration rate (eGFR) in twopatient cohorts decreased to a levelsimilar to that in control subjects andsubjects with IGT, and ACR significantlydecreased. At baseline, 86.8% and 68.4%of patients with diabetes with or withoutEHC, respectively, were not treated.Over the last 2 years from end point,IPAswere not prescribed for any patientswith EHC (Table 1).

The smoking and alcohol consumptionin the two cohorts with diabetes weresimilar between baseline and end pointand also similar between baseline and

end point in patients, control subjects,and subjects with IGT.

Comparison of CNF andNeurophysiological MeasuresBetween Baseline and End Point inPatients With Type 2 Diabetes With orWithout EHC, Control Subjects, andSubjects With IGTAll baseline CNF measures in bothcohorts with diabetes were similarbut altered compared with control sub-jects. The normalized HbA1c levels for2 years in patients with EHC im-proved all CNF measures but notCNFD (P = 0.225), with no improve-ment in those without EHC. CNFD andCNFL at end point in patients withEHC were less than those in IGT, whileother CNF measures were similarto those in subjects with IGT. AllCNF parameters but CNFD in subjectswith IGT were significantly less thanthose in control subjects (Fig. 1A–F andTable 2).

The normalized HbA1c levels in pa-tients with EHC significantly decreasedNDS, though NDS was still higher thanin control subjects and subjects withIGT. NDS in subjects with IGT was sig-nificantly higher than that in controlsubjects. All baseline neurophysiologicalmeasures in two cohorts with diabeteswere similar. The normalized HbA1c lev-els for 2 years with EHC improved re-sults of all neurophysiological tests tobe similar to those in subjects withIGT except for the median nerve am-plitude; in patients without EHC, GCimproved only VPT. Most neurophysio-logical tests in subjects with IGT wereinferior to those in control subjects(Table 2). The prevalence of retinop-athy and ETDRS retinopathy scale atend point increased insignificantly with(P = 0.183 and 0.180, respectively) orwithout (P = 0.974 and 0.317, re-spectively) EHC. EHC significantly de-creased the prevalence of neuropathyand nephropathy, while GC withoutEHC decreased only nephropathy(Table 2).

Comparison of Interval ChangesDuring Follow-up Period in CNF andNeurophysiological MeasuresBetween NGT and IGT Subgroups ofType 2 Diabetes With EHCThe improvements in NDS (mean 6SEM 21.36 6 0.20 vs. 20.68 6 0.21),

112 Impact of Normalized HbA1c on DPN Diabetes Care Volume 42, January 2019

Table

1—Anthro

pometric

andclin

icalch

aracte

risticsatb

ase

line,m

eanleve

ls,andate

ndpointin

patie

ntswith

type2diabetestre

atedwith

orwith

outE

HC,co

ntro

lsubjects,a

ndIG

Tsu

bjects

atbase

line

Patients

with

type2diab

etes

Contro

lsubjects:

baselin

eSubjects

with

IGT:

baselin

e

With

EHC

With

outEH

C

Baselin

e

Mean

levels

Endpoint

Baselin

eEndpoint

Mean

Whole

follow-up

perio

dLast

2years

from

endpoint

Number

(M/F)

38(25/13)

38(25/13)

38(25/13)

38(25/13)

38(25/13)

38(25/13)

38(25/13)

48(27/21)

34(16/18)

Age

(years)49.3

61.6

51.06

1.652.2

61.6

53.36

1.649.8

61.2

53.96

1.251.8

61.2

50.66

1.851.8

62.4

BMI(kg/m

2)28.8

60.8

25.96

0.725.8

60.7

26.26

0.7*†‡

26.86

0.8§26.7

60.9

26.16

0.922.5

60.5

22.96

0.8

Follow-upperio

d(years)

4.36

0.44.1

60.2

Duratio

nofdiab

etes(years)

4.16

0.78.2

60.8

6.06

0.910.1

61.0

SBP(m

mHg)

1526

3.4136

61.3

1346

1.6135

61.8*

1496

3.8137

61.2

|139

61.4

1326

2.4135

62.9

DBP(m

mHg)

91.46

1.679.9

60.7

78.26

0.977.9

61.2*

88.76

1.780.8

61.3*

83.56

1.3¶79.6

61.3

83.06

1.4

No.treated

with

ARB(%

)5.3

47.3*#**23.7§

57.9*12.5

11.8

CPPG

(mg/d

L)217

614.2

1206

3.0112

63.1

1166

4.8*240

614.9

1546

8.2*††

1516

4.2††

SD(m

g/dL)

30.86

2.418.1

61.4

‡‡

39.46

2.3¶CV(%

)25.4

61.8

15.86

1.0‡‡

26.36

1.6

HbA1c(%

)9.6

60.30

6.16

0.055.8

60.04

5.96

0.05*†

9.66

0.317.0

60.07*

††

7.06

0.04††

5.76

0.04**6.0

60.05

SD(%

)0.74

60.07

0.146

0.01‡‡

0.646

0.06CV(%

)12.1

61.1

2.46

0.1‡‡

9.206

0.93§

HbA1c(m

mol/m

ol)

81.66

3.443.2

60.53

40.26

0.4541.4

60.50*

†82.0

63.4

52.66

0.74*††

53.06

0.48††

39.36

0.47**41.8

60.54

75-gOGTT

FPG(m

g/dL)

1026

2.296.1

61.1

96.96

1.3PG

at120

min

(mg/d

L)152

67.0#

1076

3.0165

63.4#

LDLcholestero

l(m

mol/L)

3.666

0.153.14

60.10

3.146

0.113.32

60.13

3.866

0.173.38

60.16

|3.44

60.12

3.136

0.103.39

60.14

No.treated

with

statins(%

)2.6

5.313.2

10.518.8

14.7HDLcholestero

l(m

mol/L)

1.366

0.061.44

60.05

1.486

0.061.47

60.06

1.416

0.061.45

60.07

1.466

0.071.75

60.08

1.706

0.09Triglycerides

(mmol/L)

2.316

0.311.48

60.11

1.406

0.101.56

60.12*#

2.536

0.282.14

60.31

1.956

0.13¶0.97

60.09

‡1.47

60.18

eGFR

(mL/m

in)

88.76

3.278.7

62.6

76.66

2.874.8

62.7*

89.26

4.376.7

63.2*

79.16

2.979.6

62.0

75.86

2.3ACR(m

g/gCr)

82.96

28.037.2

612.4

21.16

6.625.8

68.1*

82.36

28.531.7

69.0§§

38.66

11.18.0

61.5

14.66

5.0

Hypoglycem

ictreatm

ent

None/IPA

/ISA/DPP4-I/d

ietalo

ne(no.)

33/2/2/3/00/0/29/8/5

0/0/27/6/526/10/8/8/0

0/2/29/26/0

Data

aremean

s6

SEMunless

otherw

iseindicated

.Data

forpatien

tswith

type2diab

eteswith

orwith

outEH

Care

from

baselin

eanden

dpoint,as

well

as(fo

rpatien

tswith

EHC)over

thewhole

follow-upperio

dorforthelast

2years

befo

reen

dpoint.Data

forcontro

lsubjects

with

NGTare

from

baselin

e,as

aredata

forsubjects

classified

asIGTby75-g

OGTT

atbaselin

e.Cr,creatin

ine;

DPP4-I,

dipep

tidyl

peptid

ase4inhibito

r;F,

female

;FP

G,fastin

gplasm

aglu

cose;M,male

;PG,plasm

aglu

cose.*P,

0.001co

mpare

dwith

base

line;†P

,0.01co

mpare

dwith

contro

lsubjects;

‡P,

0.05co

mpare

dwith

IGTsu

bjects;

§P,

0.05co

mparedwith

patie

nts

with

EHC;|P

,0.01co

mparedwith

baselin

e;¶P,

0.01co

mparedwith

patie

nts

with

EHC;#P,

0.001

comparedwith

contro

lsu

bjects;

**P,

0.001co

mparedwith

IGTsu

bjects;

††P

,0.001co

mparedwith

patie

nts

with

EHC;‡‡P

,0.001co

mparedwith

whole

follo

w-upperio

d;

§§P,

0.05co

mparedwith

baselin

e.

care.diabetesjournals.org Ishibashi and Associates 113

median nerve MCV (1.84 6 0.26 vs.0.89 6 0.25), and VPT (20.81 6 0.06vs. 20.26 6 0.16) in the NGT sub-group of type 2 diabetes with EHCwere significantly (P , 0.05) greaterthan those in the IGT subgroup.

CNFD, CNFL, CNBD, and all otherneurophysiological tests in the NGTsubgroup improved more than thosein the IGT subgroup, but the differ-ences were insignificant (P = 0.118–0.787).

Associations Between Changes inNOMs and Mean Clinical Factors, GV,and Baseline HbA1c in Type 2 Diabetes,With HbA1c Levels Normalized by EHC

The mean HbA1c levels during the wholefollow-up period were closely related

Figure 1—Representative confocal microscopic images of the corneal subbasal nerve plexus in a control subject (A) and a subject with IGT (B), atbaseline (C ) and end point (D) in a patient with type 2 diabetes without EHC, and at baseline (E) and end point (F) in a patient with EHC. Red arrows,main nerve fibers; blue arrows, branches. Linear regression betweenmean HbA1c levels and changes during follow-up period in CNFD (G), CNFL (H),CNBD (I),mediannerveMCV(J), sural nerveSCV (K), orWPT (L) inpatientswith type2diabeteswithEHC. Thedifference=value atendpoint–valueatbaseline. no, number; sec, second.

114 Impact of Normalized HbA1c on DPN Diabetes Care Volume 42, January 2019

with the improvement in CNFD, CNFL,CNBD, median nerve MCV, sural nerveSCV, and WPT (Fig. 1G–L).The multiple regression analysis in-

dicates that the changes in all CNFand neurophysiological measures byEHC were closely associated with themean HbA1c level over the wholefollow-up period. Higher SD and CVof HbA1c levels and CPPG were nega-tively associated with some CNF andneurophysiological measures. The highbaseline HbA1c levels deterioratedmany NOMs (Table 3). High DBP, LDLcholesterol, and triglycerides levelsnegatively influenced interval chan-ges in median nerve MCV, ulnar andsural nerve SCV, sural nerve ampli-tude, CPT (standard b 20.277 to20.411 and 0.322–0.336, P = 0.005–0.049), CNFD, CNFL, and beading fre-quency (standard b 20.329 to 0.588;P = 0.022–0.008).

Ability to Differentiate on the Basis ofCNF andNeurophysiological Measuresat Baseline and End Point AmongControl Subjects, Subjects With IGT,and Patients With Type 2 DiabetesWith Normalized HbA1c LevelsAt baseline and end point, with use ofCNFD and CNFL, our ability to differen-tiate between control subjects, subjectswith IGT, and patients with type 2 di-abetes with EHC was excellent. The neu-rophysiological tests clearly differentiatedcontrol subjects from patients withtype 2 diabetes. However, betweensubjects with IGT and control subjectsor patients, the sensitivity and specific-ity of CNFD and CNFL were not alwaysgood (Supplementary Table 1).

CONCLUSIONS

The intensive GC in type 2 diabetes isassociated with a reduction in microvas-cular complications (mostly albuminuria).

The strict GC may play a role in preventingand ameliorating DPN. However, in theKumamoto Study (18), the tight GC (HbA1c7.1% [54.1 mmol/mol]) prevented NCVbut not CAN decline in type 2 diabetes.In the UK Prospective Diabetes Study(UKPDS) (2), the intensive and conven-tional GC (HbA1c 7.0% vs. 7.9% [53.0 vs.62.8 mmol/mol]) had a similar effect onDPN and CAN. In the ACCORD (Action toControl Cardiovascular Risk in Diabetes)trial (3), the intensive treatment (HbA1c6.3% [45.4 mmol/mol]) prevented lossof ankle jerk and light-touch sensa-tion but increased total and CVD-relatedmortality and severe hypoglycemia.These randomized trials could notestablish the optimum GC level forpreventing the neurophysiological dete-rioration in type 2 diabetes.

When patients with recent-onsettype 1 diabetes were followed undernear-normoglycemia for 24 years, the

Table 2—CNF measures, neurophysiological tests, and microvascular complications at baseline and end point in patientswith type 2 diabetes with or without EHC and at baseline in control subjects and subjects with IGT

Patients with type 2 diabetesControlsubjects:baseline

IGT subjects:baseline

With EHC Without EHC

Baseline End point Baseline End point

CNFCNF density (no./mm2) 19.1 6 0.43* 19.4 6 0.40* 19.0 6 0.96* 18.6 6 0.96 28.7 6 0.59 25.7 6 0.77†CNF length (mm/mm2) 10.3 6 0.20*‡ 10.9 6 0.23§ 9.4 6 0.44* 9.4 6 0.40 14.3 6 0.24† 12.5 6 0.33*‡CNBD (no./mm2) 8.38 6 0.35*† 9.54 6 0.36 8.85 6 0.64* 9.25 6 0.57 12.2 6 0.42† 10.1 6 0.71|Beading frequency

(no./0.1 mm) 19.9 6 0.31*‡ 21.0 6 0.29§ 19.3 6 0.40* 19.8 6 0.25 22.4 6 0.21† 21.5 6 0.35|Bead size (mm2) 10.3 6 0.11*† 9.05 6 0.14¶ 10.4 6 0.13* 9.9 6 0.13 8.31 6 0.08† 9.48 6 0.10*

Neurophysiological testsNDS 3.9 6 0.24*† 3.1 6 0.22# 3.6 6 0.39* 3.3 6 0.33 0.4 6 0.07† 1.0 6 0.12|MCV of median nerve (m/s) 52.96 0.69*†** 54.4 6 0.67 52.2 6 0.88* 53.2 6 0.93 57.0 6 0.56 55.5 6 0.56Amplitude of median

nerve (mV) 5.29 6 0.32*†# 6.00 6 0.38 5.79 6 0.30* 6.22 6 0.44 9.48 6 0.32† 8.10 6 0.38‡††SCV of ulnar nerve (m/s) 60.7 6 0.46†| 61.3 6 0.42 59.6 6 0.65| 59.9 6 0.70 63.3 6 0.52 60.5 6 0.70|Amplitude of ulnar nerve (mV) 15.66 1.07*†** 17.7 6 1.13* 16.6 6 1.18* 17.3 6 1.13 30.7 6 1.70 20.7 6 1.51*SCV of sural nerve (m/s) 46.6 6 0.54*† 47.9 6 0.55 46.5 6 0.96* 46.4 6 0.94 50.5 6 0.46‡ 46.2 6 0.59*Amplitude of sural nerve (mV) 9.67 6 0.53*† 10.8 6 0.61* 9.61 6 0.52* 9.38 6 0.54 17.5 6 0.69 10.6 6 0.59*VPT (microns/120 c/s) 2.94 6 0.21††† 2.48 6 0.19 3.42 6 0.24‡‡ 3.15 6 0.27 2.27 6 0.17 2.64 6 0.32CVR-R (%) 3.20 6 0.17†| 3.48 6 0.19 3.24 6 0.20†† 3.42 6 0.24 4.10 6 0.23 3.45 6 0.18WPT (W/m2) 2618 6 27.6*† 2535 6 39.9†† 2568 6 18.8†† 2567 6 19.5 2486 6 11.4 2545 6 18.4††CPT (W/m2) 528 6 22.6†| 481 6 23.0§§ 525 6 17.2 536 6 15.9 443 6 10.5 531 6 17.8*

Microvascular complicationsPrevalence of retinopathy (%) 7.9 13.2 21.1 23.7ETDRS retinopathy scale 0.13 6 0.08 0.21 6 0.09 0.34 6 0.11 0.39 6 0.12Prevalence of

nephropathy (%) 34.2‡‡ 18.4 31.6§§ 13.2Prevalence of neuropathy (%) 13.2‡ 2.6 18.4 10.5

Data are means 6 SEM. Data are for patients with type 2 diabetes with or without EHC at baseline and end point, control subjects with NGT,and subjects with IGT classified by 75-g OGTT at baseline. *P , 0.001 compared with control subjects; †P , 0.001 compared with end point inpatients with EHC; ‡P , 0.01 compared with end point in patients with EHC; §P , 0.01 compared with patients without EHC; |P , 0.01compared with control subjects; ¶P, 0.001 compared with patients without EHC; #P, 0.001 compared with IGT subjects; **P, 0.05 comparedwith IGT subjects; ††P , 0.05 compared with control subjects; ‡‡P , 0.05 compared with end point in patients with EHC; §§P , 0.05 comparedwith end point in patients without EHC. c, cycle.

care.diabetesjournals.org Ishibashi and Associates 115

nerve function declines were com-pletely prevented (4). Therefore, thenear-normoglycemia is prerequisite forpreventing the neurophysiological dete-rioration. We recently showed that allneurophysiological and some CNF meas-ures are improved by subnormal HbA1clevels (HbA1c 6.5% [47.5 mmol/mol])over 4 years in patients with poorlycontrolled type 2 diabetes, while somenerve functions in patients with conven-tional treatment (mean HbA1c 7.2%[55.6 mmol/mol]) deteriorated (19).Therefore, standard diabetes care(HbA1c $7.0% [53.0 mmol/mol]) is notbeneficial for DPN. In the current study,the treatment without EHC improvedonly VPT. Because of potential severehypoglycemia (3), the impact of long-term normalized HbA1c levels on improv-ing neuropathy outcomes in type 2diabetes had never been reported.

Tight GC needs to be maintained for.3–5 years to yield benefit (20). Thecurrent study followed patients withpoorly controlled type 2 diabetes for.4 years on average, and mean HbA1clevels in patients with EHC over thewhole follow-up period and last 2 yearswere 6.1% (43.2 mmol/mol) and 5.8%(39.9 mmol/mol), respectively. For thelast 2 years from end point IPAs werenot prescribed, and no hypoglycemic epi-sodes were reported. All neurophysi-ological tests and most CNF parameterswere improved significantly by normaliz-ing HbA1c levels for 2 years. The resultsshowed that the mean HbA1c levels overthe whole follow-up period played a keyrole in improving NOMs. The high baselineHbA1c levels compromised most NOMs asa negative metabolic memory (21,22).

GV may confer additional risk for DPNindependent of HbA1c levels (23,24). Inthe cross-sectional study in type 2 di-abetes, GV estimated by continuousglucose monitoring (CGM) had a closerelationship with DPN development(25). We assessed the GV using SD andCV ofmonthly measured CPPG and HbA1clevels over 4.3 years. The largeGV compro-mised some NOMs. SD and CV of CPPGcompromised most CNF parameters, sug-gesting that daily glucose fluctuation isharmful on small nerve fibers (SNFs). GVparameters obtained by CGM had beenused in a cross-sectional study associatingGVwith DPN (26). For long-term follow-upstudy, repeated CGM is required to esti-mate mean GV, but it is impractical.

Table

3—Correlationsbetw

eench

angesin

NOMsbyGC

withEHC

andmeanleve

lsofHbA1candCPPG,theirva

riability,

andbas

elineHbA1cleve

lsin

patients

withtype2

diabeteswithEHC

Interval

changesin

NOMs

ParametersofHbA1clevels

ParametersofCPP

G

Mean

SDCV

Atbaseline

Mean

SDCV

Standardb

PStandardb

PStandardb

PStandardb

PStandardb

PStandardb

PStandardb

P

CNF Den

sity

20.519

0.002

20.21

00.19

020.162

0.31

420.474

0.021

20.11

30.48

520.274

0.074

20.261

0.097

Length

20.664

<0.001

20.43

00.02

120.415

0.01

520.518

0.019

20.39

90.023

20.454

0.00

720.392

0.02

5Branch

den

sity

20.577

0.003

20.24

90.19

220.191

0.32

020.493

0.020

20.03

40.86

120.357

0.04

820.392

0.03

3Beadingfreq

uen

cy20.396

0.017

20.18

30.26

120.151

0.35

520.328

0.14

60.100

0.54

520.305

0.04

920.319

0.04

4Beadsize

0.471

0.007

0.368

0.03

00.513

0.00

10.605

0.003

0.295

0.08

70.502

0.00

10.515

0.00

1

Nerve

functions

MCVofmed

iannerve

20.481

0.002

20.21

50.15

720.153

0.31

920.487

0.002

20.11

80.44

420.111

0.459

20.059

0.702

Amplitudeofmed

iannerve

20.534

0.006

20.09

80.61

520.046

0.81

420.308

0.16

320.00

20.99

420.069

0.717

20.082

0.674

SCVofulnar

nerve

20.626

<0.001

20.39

60.01

220.626

<0.001

20.709

0.001

20.36

20.019

20.378

0.01

220.340

0.02

9Amplitudeofulnar

nerve

20.460

0.016

20.13

80.46

620.090

0.63

520.409

0.06

70.033

0.86

220.067

0.718

20.068

0.717

SCVofsuralnerve

20.474

0.001

20.38

20.00

420.330

0.01

620.733

<0.001

20.15

10.28

920.196

0.139

20.179

0.189

Amplitudeofsuralnerve

20.526

0.006

20.23

00.24

420.181

0.34

220.586

0.005

20.21

20.26

520.197

0.287

20.161

0.398

VPT

0.502

0.004

0.291

0.08

50.238

0.16

10.593

0.004

20.22

70.19

40.202

0.225

0.118

0.493

CVR-R

20.498

0.005

20.28

20.10

420.240

0.16

920.465

0.029

20.11

60.50

620.127

0.461

20.101

0.565

WPT

20.520

0.004

20.42

50.01

320.379

0.02

820.767

<0.001

20.37

10.019

20.351

0.02

520.266

0.103

CPT

0.467

0.004

0.407

0.00

90.361

0.02

10.749

<0.001

0.142

0.38

40.254

0.100

0.245

0.121

Statistically

significantcorrelationsappearin

boldface

type.

116 Impact of Normalized HbA1c on DPN Diabetes Care Volume 42, January 2019

Normal HbA1c levels donot necessarilymean normoglycemia. IGT may play animportant role in inducing neuropathy(9,27) and influences NOMs more pro-foundly than impaired fasting glucose(28). Therefore, determining glucosetolerance category (GTC) by OGTT inpatients with normal HbA1c levels afterEHC is required for assessing the influ-ence of remaining glucose intolerance onNOMs. However, OGTT had never beenincluded in clinical studies assessing thebenefit of EHC for DPN. In the currentstudy, patients with poorly controlledtype 2 diabetes returned to IGT afterEHC, while 28.9% of patients returned toNGT. The interval changes in NOMs byEHC were compared between NGT andIGT subgroups. The improvementsbyEHCinNDS,mediannerveMCV, and VPT in theNGT subgroup were significantly greaterthan in the IGT subgroup. Although in-significant, most other NOMs in the NGTsubgroup improved more than those inthe IGT subgroup, suggesting that re-maining IGT after normalizing HbA1c lev-els may compromise NOMs. The reversalof diabetes plays a role in improvingdiabetes complications. The improve-ment of NOMs in this study was signif-icant, but the changes were small. Thismight be due to remaining IGT afternormalized HbA1c levels, negative in-fluence of GV, and metabolic memoryof hyperglycemiador perhaps longerfollow-up period is required.As the SNFs are most likely to respond

to interventions for DPN (29), the mor-phological and functional SNF measuresare essential for evaluation of the benefitof normalized HbA1c levels on NOMs. CNFvisualized by corneal confocal micros-copy (CCM), predominantly SNF (30),can quantify SNF pathology in DPN(31,32). Subjects having SNF neuropathyshow a significant reduction of intra-epidermal nerve fiber density and CNFmeasures (9). In subjects with IGT,CCM reveals dynamic CNF changes re-lated to changing GTC; CNFD and CNFLincrease in IGT subjects returning toNGT,while CNFL in subjects developing type 2diabetes reduces (33). In type 1 diabetesin the context of a normal HbA1c level(5.9% [41.0 mmol/mol]), CNFD, CNFL,and CNBD were improved in 12 monthsafter SPK (5). However, this study did notdetermine the GTC after SPK.In the current study, EHC improved

all CNF measures apart from CNFD, with

no improvement in those without EHC,indicating that the normalized HbA1clevels are more beneficial than standardcare. In subjects with IGT, most CNFmeasures were altered compared withNGT subjects.

Normalizing HbA1c levels for 2 yearsimproved most neurophysiological dys-functions to levels similar to those ofsubjects with IGT, while standard careimproved only VPT, indicating the supe-riority of normalized HbA1c level by EHCcompared with standard care for ame-liorating neurophysiological dysfunc-tions. IGT impaired NDS and mostneurophysiological functions comparedwith results in control subjects with NGT.

Because functions of SNFs as well aslarge nerve fibers were impaired in sub-jects with IGT compared with NGT con-trol subjects, there was no preferentialimpairment of SNF functions in IGT.However, we performed a single OGTTat end point in patients with EHC and atbaseline in control subjects and in sub-jects with IGT. There are issues concern-ing the reproducibility and reliability of asingleOGTT in establishingGTCs (34). It isnoteworthy that the reproducibility ofIGT is lower than for other GTCs (35)which highlights a major limitation ofall studies including ours relating GTCsto impaired NOMs (35). Furthermore,we did not follow up control subjectsor subjects with IGT. Therefore, wecould not compare the time-dependentchanges in NOMs among patients withdiabetes, control subjects, and subjectswith IGT, which might function to misleadus in our attempt to draw conclusions ofthis study.

Although chronic hyperglycemiaclearly plays a causative role in develop-ment of DPN in type 2 diabetes, meta-bolic syndrome components are likely tocause neuropathy (36). Obesity and glu-cose intolerance have a definite role inneuropathy, whereas hypertension anddyslipidemia have equivocal results. Inthis study, BMI, SBP, DBP, and triglycer-ides decreased after GC. The improve-ments in BMI, DBP, and dyslipidemiaduring follow-up were significantly asso-ciated with the improvement of someCNF and neurophysiological parameters.Therefore, besides the strict GC andsmaller GV, the metabolic syndromecomponents should be controlled forameliorating DPN even in the contextof a normal HbA1c level. Smoking and

alcohol consumption, which may influ-ence DPN (36), had no impact on NOMsin patients with diabetes in this study.The normalized HbA1c levels for 2 yearsdecreased nephropathy, while reti-nopathy increased insignificantly. Theinitial extreme hyperglycemia and sub-stantial HbA1c reduction might developthe retinopathy (37). Although angioten-sin receptor blockers (ARBs) may pro-tect against DPN, ARBs did not improveany NOMs (standard b = 0.011 to20.347; P = 0.918–0.062).

Assessment of the performance andvalidity of CCM in longitudinal studies intype 2 diabetes is not an area of studythat has investigated. The current studyis the only study that has measured themain DPN risk factors, including CPPG,HbA1c levels, blood pressure, and BMI,monthly, and the lipid profile, eGFR, andACR (every 3 months). Therefore, themean clinical factors influencing NOMsare representative; the benefits of nor-malizedHbA1c levels and improvement inother risk factors to NOMs were reliablyevaluated. We estimated the morpho-logical and functional SNF measures.CNFs at baseline in patients and in sub-jects with IGT were already altered. Thenormalized HbA1c levels for 2 years im-proved all neurophysiological tests ofSNFs and most CNF measures.

Our study had many limitations.Firstly, we assessed only baseline andend point NOMs. Ideally, these exami-nations should be performed annually forassessing trends in neuropathy changes.Secondly, we used SD and CV of CPPG andHbA1c levels as GV parameters; however,GV can be assessed by several methods,and there is little consensus regardingthe optimal method. Thirdly, patientswith EHC at end point, control subjects,and subjects with IGT at baseline un-derwent a single OGTT. Therefore, thereproducibility of the OGTT was poor.Repeated OGTTs are desirable for asso-ciation of GTCs with neuropathy prog-ress. Fourthly, the prospective follow-upstudyof a large number of patients over alonger follow-up period is mandatory toreinforce the present results and estab-lish the relationship between the devel-opment of neuropathy and IGT. Finally,the patients with type 2 diabetes in thecurrent study had a relatively short di-abetes duration. Because the duration ofdiabetes is one of the major risk factorsfor DPN and other complications, and

care.diabetesjournals.org Ishibashi and Associates 117

severe DPN may be less amenable tointervention, we could not extend thepresent results to patients with longerduration or advanced neuropathy.In conclusion, patients with poorly

controlled type 2 diabetes returned toIGT by normalizing HbA1c level for 2 yearswithout hypoglycemia. All neurophysio-logical tests and most CNF measuresimproved to IGT levels. However, stan-dard care without EHC improved onlyVPT despite the same hypoglycemicstrategies. This is due to a big differencein body weight reduction during follow-up. The high GV, negative metabolicmemory, and metabolic syndrome com-ponents are harmful tomost NOMs evenin the context of normal HbA1c levels.Along with normal HbA1c levels by EHCwithout hypoglycemia, small GV andcontrol of metabolic syndrome compo-nents are indicated for preventing andameliorating DPN. The normalized HbA1clevels are more effective than standardcare for preventing the development ofneuropathy but not retinopathy.

Funding. This study received no financialsupport.Duality of Interest. No potential conflicts ofinterest relevant to this article were reported.Author Contributions. F.I. designed the study,researched data, and wrote the entire manu-script.M.Tan. performed CCMexaminations andneurophysiological tests. A.K. and H.U. gatheredthe clinical and laboratory data and statisticallyanalyzed all data. M.Tav. advised on the statis-tical analysis, interpreted the results, and re-viewed and revised the whole manuscript. F.I.and M.Tav. are the guarantors of this work, and,as such, had full access to all data in the study andtake responsibility for the integrity of the dataand the accuracy of the data analysis and in-terpretation.

References1. Ang L, Jaiswal M, Martin C, Pop-Busui R. Glucosecontrol and diabetic neuropathy: lessons from re-cent large clinical trials. Curr Diab Rep 2014;14:5282. UK Prospective Diabetes Study (UKPDS)Group. Intensive blood-glucose control withsulphonylureas or insulin compared with con-ventional treatment and risk of complications inpatients with type 2 diabetes (UKPDS 33) [pub-lished correction appears in Lancet 1999;354:602]. Lancet 1998;352:837–8533. Ismail-Beigi F, Craven T, Banerji MA, et al.;ACCORDTrialGroup. Effect of intensive treatmentof hyperglycaemia on microvascular outcomes intype 2 diabetes: an analysis of the ACCORDrandomised trial. Lancet 2010;376:419–4304. Ziegler D, Behler M, Schroers-Teuber M,Roden M. Near-normoglycaemia and develop-ment of neuropathy: a 24-year prospective study

from diagnosis of type 1 diabetes. BMJ Open2015;5:e0065595. Tavakoli M, Mitu-Pretorian M, PetropoulosIN, et al. Corneal confocal microscopy detectsearly nerve regeneration in diabetic neuropathyafter simultaneous pancreas and kidney trans-plantation. Diabetes 2013;62:254–2606. Jin HY, Lee KA, Park TS. The impact of glycemicvariability on diabetic peripheral neuropathy.Endocrine 2016;53:643–6487. Pop-Busui R, Herman WH, Feldman EL, et al.;DCCT/EDIC Research Group. DCCT and EDICstudies in type 1 diabetes: lessons for diabeticneuropathy regarding metabolic memory andnatural history. Curr Diab Rep 2010;10:276–2828. Chalmers J, CooperME.UKPDS and the legacyeffect. N Engl J Med 2008;359:1618–16209. Asghar O, Petropoulos IN, Alam U, et al.Corneal confocalmicroscopydetects neuropathyin subjects with impaired glucose tolerance.Diabetes Care 2014;37:2643–264610. Pop-Busui R, Lu J, Brooks MM, et al.; BARI2D Study Group. Impact of glycemic controlstrategies on the progression of diabetic pe-ripheral neuropathy in the Bypass AngioplastyRevascularization Investigation 2 Diabetes (BARI2D) cohort. Diabetes Care 2013;36:3208–321511. Wilkinson CP, Ferris FL III, Klein RE, et al.;Global Diabetic Retinopathy Project Group. Pro-posed international clinical diabetic retinopathyand diabetic macular edema disease severityscales. Ophthalmology 2003;110:1677–168212. Molitch ME, DeFronzo RA, Franz MJ, et al.;American Diabetes Association. Nephropathy indiabetes. Diabetes Care 2004;27(Suppl. 1):S79–S8313. YoungMJ, BoultonAJ,MacLeodAF,WilliamsDR, Sonksen PH. A multicentre study of theprevalence of diabetic peripheral neuropathyin the United Kingdomhospital clinic population.Diabetologia 1993;36:150–15414. Tesfaye S, BoultonAJ, Dyck PJ, et al.; TorontoDiabetic Neuropathy Expert Group. Diabeticneuropathies: update on definitions, diagnosticcriteria, estimation of severity, and treatments.Diabetes Care 2010;33:2285–229315. Ishibashi F, Okino M, Ishibashi M, et al.Corneal nerve fiber pathology in Japanesetype 1 diabetic patients and its correlationwith antecedent glycemic control and bloodpressure. J Diabetes Investig 2012;3:191–19816. Tavakoli M, Malik RA. Corneal confocalmicroscopy: a novel non-invasive technique toquantify small fibre pathology in peripheralneuropathies. J Vis Exp 201117. Ishibashi F, Kojima R, TaniguchiM, Kosaka A,Uetake H, Tavakoli M. The expanded bead sizeof corneal C-nerve fibers visualized by cornealconfocal microscopy is associated with slowconduction velocity of the peripheral nervesin patients with type 2 diabetes mellitus. J Di-abetes Res 2016;2016:365345918. Ohkubo Y, Kishikawa H, Araki E, et al. In-tensive insulin therapy prevents the progressionof diabetic microvascular complications in Jap-anese patients with non-insulin-dependent di-abetes mellitus: a randomized prospective 6-yearstudy. Diabetes Res Clin Pract 1995;28:103–11719. Ishibashi F, Tavakoli M. Impact of normo-glycemia in reducing microvascular complicationsinpatientswith type2diabetes: a follow-up study.Front Endocrinol (Lausanne) 2018;9:52

20. Giorgino F, Home PD, Tuomilehto J. Glucosecontrol and vascular outcomes in type 2 diabe-tes: is the picture clear? Diabetes Care 2016;39(Suppl. 2):S187–S19521. Duckworth W, Abraira C, Moritz T, et al.;VADT Investigators. Glucose control and vascu-lar complications in veterans with type 2 di-abetes. N Engl J Med 2009;360:129–13922. Intine RV, Olsen AS, SarrasMP Jr. A zebrafishmodel of diabetes mellitus and metabolic mem-ory. J Vis Exp 2013;72:e5023223. Gorst C, Kwok CS, Aslam S, et al. Long-termglycemicvariabilityandriskofadverseoutcomes:a systematic review and meta-analysis. DiabetesCare 2015;38:2354–236924. Yang CP, Li CI, Liu CS, et al. Variability of fastingplasma glucose increased risks of diabetic poly-neuropathy in T2DM. Neurology 2017;88:944–95125. Xu F, Zhao LH, Su JB, et al. The relationshipbetween glycemic variability and diabetic peripheralneuropathy in type 2 diabetes with well-controlledHbA1c. Diabetol Metab Syndr 2014;6:13926. DeVries JH. Glucose variability: where it isimportant and how tomeasure it. Diabetes 2013;62:1405–140827. Smith AG, Russell J, Feldman EL, et al. Life-style intervention for pre-diabetic neuropathy.Diabetes Care 2006;29:1294–129928. DimovaR, TankovaT, GuergueltchevaV, et al.Risk factors for autonomic and somatic nervedysfunction in different stages of glucose toler-ance. J Diabetes Complications 2017;31:537–54329. Malik RA, Veves A, Tesfaye S, et al.; TorontoConsensus Panel on Diabetic Neuropathy. Smallfibre neuropathy: role in thediagnosis of diabeticsensorimotor polyneuropathy. Diabetes MetabRes Rev 2011;27:678–68430. Muller LJ, Marfurt CF, Kruse F, Tervo TMT.Cornealnerves: structure, contents and function.Exp Eye Res 2003;76:521–54231. Quattrini C, Tavakoli M, Jeziorska M, et al.Surrogate markers of small fiber damage inhuman diabetic neuropathy. Diabetes 2007;56:2148–215432. Chen X, Graham J, Dabbah MA, et al. Smallnerve fiber quantification in the diagnosis ofdiabetic sensorimotor polyneuropathy: compar-ing corneal confocal microscopy with intraepi-dermal nerve fiber density. Diabetes Care 2015;38:1138–114433. Azmi S, Ferdousi M, Petropoulos IN, et al.Corneal confocal microscopy identifies small-fiber neuropathy in subjects with impaired glu-cose tolerance who develop type 2 diabetes.Diabetes Care 2015;38:1502–150834. Rajabally YA. Neuropathy and impairedglucose tolerance: an updated review of theevidence. Acta Neurol Scand 2011;124:1–835. Balion CM, Raina PS, Gerstein HC, et al.Reproducibility of impaired glucose tolerance(IGT) and impaired fasting glucose (IFG) classi-fication: a systematic review. Clin Chem LabMed2007;45:1180–118536. Grisold A, Callaghan BC, Feldman EL. Medi-ators of diabetic neuropathy: is hyperglycemiathe only culprit? Curr Opin Endocrinol DiabetesObes 2017;24:103–11137. ShurterA,GenterP,OuyangD, IppE.Euglycemicprogression: worsening of diabetic retinopathy inpoorly controlled type 2 diabetes in minorities.Diabetes Res Clin Pract 2013;100:362–367

118 Impact of Normalized HbA1c on DPN Diabetes Care Volume 42, January 2019